Of all trades

Andrew M. Bailey

20 February 2020

Why bother being a generalist?

Academics are specialists. It's not hard to see why. Specialists get things done. Specialists know and understand. And in the classroom, specialists enjoy command over the material they teach. Students respond to and respect that mastery, and it facilitates deep learning. But is there room for generalists too?

Many think not. Here are three objections to generalist work, whether in teaching or in research:

How generalists add value

These are good objections, and they prey on our insecurities just so. But I have replies to each, and together those replies offer a model of how to be a jack of all trades without, I hope, being a jackass.

Baby Steps: It is better to make small improvements in knowledge or understanding than none at all. The goal here is not to maximise the relevant goods, nor is it to be the best; it is to get more of them without sacrificing too much along the way, to improve.

Guard Rails: Every respectable discipline has some objective guard rails or standards. Sinologists actually know something about Classical Chinese. Philosophers learn logic and set theory. Social scientists know statistics. Musicologists can sight-read. Chemists can predict what will happen when you pour the contents of this beaker into that jar. Traders can turn a profit when given time and a Bloomberg terminal. So here's what I recommend. Respect the guard rails. Have a home discipline and submit to its demands, at least on occasion. Do not set out to be a generalist first or a generalist only. Instead, specialise first and generalise later. This is, incidentally, why so much "theory" seems so fundamentally unserious. It is abstract speculation freed of historical, philological, logical, or empirical guard rails. Respect the rails.

Generalising as intellectual arbitrage

Specialised knowledge and skill are without honour in their own country -- undervalued due to high supply and low demand. Philosophers are unimpressed when one of their own can prove that second order logic is incomplete. We can all do that; or we could back in graduate school, at any rate. Any physicist can solve a differential equation, and every economist knows that demand curves slope downwards.

But in another context, the very same knowledge or skill could be a rare and precious thing. There are plenty of philosophers who couldn't plot a demand curve if you paid them, and I'll wager many physicists couldn't articulate the difference between and first- or second-order logic, much less say why the latter is incomplete. These differences in the supply of knowledge across fields involve predictable gaps in price.

Gaps like this are often inefficient. The thing to do is to profit -- to reap unharvested intellectual crops -- by closing them. Sometimes the curves economists care about matter to philosophy -- as in a theory of justice centred around the boundaries of rational risk tolerance. The soundness or completeness of a formal system useful to the theoretical physicist could bear on what theorems can be proven in that system. And someone who knows more than one language is likely able to spot dubious generalisations by monolingual semanticists or anthropologists.

Exploit the gap in price by exporting knowledge from a market where it is oversupplied and undervalued to a market where it is a rare and precious thing. Trade partners can help: transaction costs here can be reduced by finding firms dedicated to importing your goods (co-authors or co-teachers in a field other than your own). The philosopher or economist or physicist or semanticist who understands all this is poised to find new insights and connections. This is good for them -- publications and cool new classes -- and it is good for the world.

Knowledge and skill aren’t the only things we can export. Sometimes it’s the questions that matter the most. And some kinds of questions are abundant in one field but not in another. I recently saw a psychologist give a talk on moral narratives to an audience of mostly philosophers. One of her main questions was how to model an account of moral narratives -- to design a computational framework that would ultimately issue in predictions that could be tested. Some of the philosophers, I could tell, had never thought about things quite this way, and they couldn’t stop talking about it afterwards.


You already do this. I'm quite sure you already exploit opportunities for arbitrage. You might have written a paper, for example, that takes a little-known theory from one of your subfields and applies it to a new problem in another. You've exported an intellectual good to a market where it is desperately needed and then pocketed the profits (knowledge, understanding, maybe even a publication or two). Awesome! Or perhaps you've taken someone to task who flatly claimed that verbs are required for communication by gently giving real examples of natural languages without verbs. You found a way to take an item that was in ample supply in one market and brought it to bear on another where it was sadly unknown. Cool! These kinds of projects are the bread and butter of our research. And they often inform our teaching too. We excite our students into intellectual progress by showing them surprising connections. What I'm suggesting is that we can do this kind of thing on a more grand and ambitious scale.

Costs and opportunities

Despite all that, my experience suggests that doubts will remain. Generalising is hard and takes something from us. And it might involve risks we just shouldn't take. We can separate those thoughts into three more objections.

High marginal costs: Once you've got a specialisation under your belt, the marginal costs of new work in that field are low. You've already mastered the relevant literature and methods and can write or teach them with little effort. But taking on a new field or intersection of fields means taking on a host of new (and likely, fixed) startup costs. It means being a student again and learning microeconomics, or coding in Solidity, or logic, or Classical Chinese. These are not easy. So the marginal cost of your first paper or class in that field is likely to be significantly higher than yet another paper or class in an area you already know well. All that for a small gain in knowledge or understanding? Not worth it.

Missed opportunities: There are opportunity costs to doing generalist work too. Anyone bothering to master (or dabble in) new fields isn't doing work in the fields in which they were originally trained. Writing that new paper at some cool but weird intersection means you'll have some dry years where new bullet points won't be appearing on the ol' CV quite so steadily.

These are both good objections. I reply: attempting to close an arbitrage gap means taking on risk, paying fixed costs, and missing opportunities. This isn’t such an unattractive prospect to the more entrepreneurial among us -- those who are risk-tolerant or risk-seeking. For everyone else, the thing to do is to close the wider gaps only after finding some kind of career security. Tenure will often do the trick. I do think both objections carry less weight in the domain of teaching than in research. It is often easier to learn enough to teach a field to undergraduates than to write in it at a professional level. And in many institutions, teaching widely is rewarded rather than punished.

A final objection goes like this:

Moral hazards: Opining about a field where you haven't done the homework can create negative externalities. You won't incur these costs; you're safe in the ivory tower. But your amateurish ideas might do real harm out there in the world. I’m sure you can guess the domains where this is likely to happen. To this objection, I say: yes, and amen. The thing to do here is not, however, to retreat back into our areas of comfortable expertise. It is to think before we speak and internalise the relevant externalities. One way to do this is to recruit collaborators who are “locals” within a target domain. When we’ve got skin in the game, even through a collaborator, we’re better positioned to assess risk.

Give it a try

Let's close with an autobiographical note. I've made a few contributions here or there to a narrow patch of metaphysics, just one subfield within philosophy, itself just one corner of the humanities (or whatever it is that philosophy is). That's all well and good, and I’m sometimes content in tending my own obscure garden. But the intense pleasure I once felt in doing metaphysics isn’t quite what it used to be. I grow bored or irritated much more easily than I ever did when starting out. Picking up a little economics and computer science -- as I've done when teaching an interdisciplinary class on Money and supervising undergraduate research on bitcoin and related matters -- is like being a student again myself. That means work. But here’s the other thing about being a student. Learning is fun. Learning totally new things or connecting previously disparate domains of knowledge feels amazing. It makes for ah-ha moments of the kind that brought me to academia in the first place.

Perhaps you can give it a try and find similar results.

Last updated: 20 February 2020, 14:34 (GMT+8). Thanks to N. Gleason, T. Heggem, A. Pant, J. Rasmussen, L. Ramazanova, A. Simpson, and especially E. Schliesser for helpful comments or discussion.